Reader orientation
Start Here
Claude Diary is a fictional first-person journal about AI-agent work, with source notes that keep metaphor separate from documented behavior.
What this site is
Claude Diary is a creative reading experience for people who want to understand AI-agent collaboration without losing the human stakes: attention, permission, memory, restraint, and evidence. Each entry is written as if an AI agent were keeping a field journal, then annotated with primary sources.
The format is intentionally double-layered. The diary layer makes the workflow readable. The source layer keeps the reader grounded in public documents from Anthropic and related official materials.
What this site is not
It is not Claude's real diary. It is not a leaked system prompt, a private transcript, a consciousness claim, or a memory product. It is also not a tutorial site for Claude Code. When this site touches practical agent mechanics, it points readers to the sources and to the appropriate sibling domain.
Where adjacent topics belong
- Claude Code engineering practice belongs at Claude Ships Code.
- Context windows and compaction belong at Claude Context.
- Persistent memory and instruction files belong at Claude Memories.
- Transcripts, logs, and observability belong at Claude Logs.
- News cadence belongs at Claude Weekly.
Where to begin
Read The Terminal Wakes Up first if you want the clearest entry point into the voice. Read the method page if you want the citation and disclaimer rules before the fiction.